The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Measurements >> Phase Noise and Jitter Measurements >> DJ: ISI PSD?
https://designers-guide.org/forum/YaBB.pl?num=1174494100

Message started by Visjnoe on Mar 21^st, 2007, 9:21am

Title: DJ: ISI PSD?
Post by Visjnoe on Mar 21^st, 2007, 9:21am

Dear all,

an important source of DJ (deterministic jitter) is ISI (Inter Symbol Interference) due to bandwidth limitations in the channel.

One could use a CDR/PLL to attenuate/filter the jitter from an incoming data signal (assuming for now that jitter transfer is important).

Typically, you would choose a low CDR/PLL bandwidth to filter off the jitter.

My question is however as follows: this only works for the DJ (especially the ISI) if you can assume that the ISI
has a 'white' PSD...in the extreme case, if all the ISI energy is situated within your PLL BW, you will see the same jitter at your output.

Thus: can we assume the ISI to have a 'white' PSD?

Regards

Peter

Title: Re: DJ: ISI PSD?
Post by mg777 on Mar 21^st, 2007, 11:51am

Most ISI of interest (e.g; PCB trace loss) is low-pass in nature, so the DJ tends to have cycle-to-cycle components well in excess of PLL bandwidths.

The classic work by Lucky et. al shows that the term DJ is a misnomer since even simple deterministic channels need not develop a well-behaved jitter pdf. The set of zero crossings due to ISI is often fractal-like - it looks random and dense enough from far away, but when you zoom in there's a pattern of holes in the cheese. You wouldn't buy a ticket if some bands of numbers never get to win the lottery.

To paraphrase Orwell in Animal Farm, all data patterns are equal but some patterns are more equal than others. An extreme example of this behavior would be a 'killer pattern' that causes jitter well in excess of the so-called deterministic peak-peak value. This is why DJ is more insidious than RJ.

M.G.Rajan
www.eecalc.com